Assessment of ice impact load threshold exceedance in the propulsion shaft of an ice-faring vessel via Bayesian inversion

Ice-induced impact loads on the propeller blades of vessels operating in polar waters represent a notable hazard to ship propulsion systems. Industry guidelines determine the maximum allowable ice-induced propeller moments, but these loads are not practical to measure directly. Recent studies have implemented deterministic inversion methods to determine these impact-induced moments from torque measurements on the propulsion shaft. This study considers this inversion problem from a stochastic perspective. Bayesian inversion is used, first, as a means of determining the optimal regularisation parameters, and second, as an avenue to explore the contributions to uncertainty in the inverted ice loading values due to the linear inversion model. The method is implemented for two sets of shaft-line measurements recorded on the SA Agulhas II, a polar class PC-5 research and supply vessel, via a lumped-mass model for the forward response signal. Using more sophisticated simulation results as prior information for the model uncertainty, it is shown that inverted ice-induced propeller moments around 20% below the industry guideline maximum loading of 1009.9 kN m correspond to a 1% probability of ice-induced moments exceeding this limit. This finding is significant, given the number of propeller impact events for the SA Agulhas II that approached this limit during recent voyages, and highlights the need for considering design specifications for propulsion systems in ice-faring vessels from a risk perspective.